Apple A8 and A8X

Some rumors that the iPads to be unveiled next week will have 2 GB of RAM, unlike the 1 GB in the iPhone 6 and 6 Plus.
 
Some rumors that the iPads to be unveiled next week will have 2 GB of RAM, unlike the 1 GB in the iPhone 6 and 6 Plus.

Some have found evidences of 3x assets for iPad in iOS 8.1, so maybe there'll be a new iPad with higher resolution.
 
And WSJ reported that a bigger iPad will be delayed to next year because the suppliers are concentrating on iPhone 6 production.
 
And WSJ reported that a bigger iPad will be delayed to next year because the suppliers are concentrating on iPhone 6 production.

I was actually surprised to read that everyone was expecting it all of the sudden on the 16th of October. Early reports for a >12" "PRO" tablet pointed at early next year quite a while ago. So it's either truly delayed or happygomerry rumor mongerers don't have a clue (which wouldn't be much of a surprise either).
 
I copied your reply from the Tegra thread since its getting too much OT there. I don't know where the GX6450s are clocked in A80 but I could imagine that in the 6 Plus the GPU goes slightly over 500MHz. They could go for 700 or even more MHz, but to stay at a similar to iPhone/A8 frequency with a GX6650 sounds way more reasonable from a power budget to me.


Well, it's back to the old question as to how they balance the R&D / production / inventory / chip size costs of a new SOC against the benefits it provides.

I think they did a A6/A6X because they just couldn't come up with a compromised SOC that would do both smartphone & tablet, remembering at the time that the tablet pixels/smatphone pixels ratio was a lot bigger then, than it is today.

A7 was for both devices, with just a modest tweak.

The range of pixel counts across device is not as wide for this current cycle, so it is more likely, IMO, that it'll be one size fits all, especially if the new ipad air doesn't have a more dense pixel count. The power saving from A7-A8 might be given to GPU, assuming most other internals don't change much.

I assume a 12" ipad Air Pro would need a different soc. If a new gen ipad air is getting a denser screen, then that would change the decision making.
 
I was actually surprised to read that everyone was expecting it all of the sudden on the 16th of October. Early reports for a >12" "PRO" tablet pointed at early next year quite a while ago. So it's either truly delayed or happygomerry rumor mongerers don't have a clue (which wouldn't be much of a surprise either).
Well the earlier rumors in 2013 pointed to an 2014 introduction of an iPad "Pro," and I think that it wasn't until the last half year or so that rumors directly pointing to 2015 showed up.

The range of pixel counts across device is not as wide for this current cycle, so it is more likely, IMO, that it'll be one size fits all, especially if the new ipad air doesn't have a more dense pixel count. The power saving from A7-A8 might be given to GPU, assuming most other internals don't change much.

I assume a 12" ipad Air Pro would need a different soc. If a new gen ipad air is getting a denser screen, then that would change the decision making.
Suppose an 12"-13" iPad "Pro" is coming before 2015 H2, the iPad Air is updated next week, no further iPad Air updates will occur until 2015 H2, and Apple thinks the A8 is insufficient for a 3072x2304 display. I wonder if part of the reason for the 3x display on the iPad Air is because Apple decided to use an A8X* in the iPad "Pro," and to meet required volume for the A8X, put it in the iPad Air as well. Then the situation of Apple choosing to release an even thinner 2x Retina iPad Air with an A8 instead just isn't in the cards. (The opposite situation isn't the same, as the iPad Air sells enough such that Apple could release a 3x iPad Air with A8X whether or not the iPad "Pro" exists.)

* A8X here refers to any "A8-" branded SoC with more CPU or GPU cores than the vanilla A8.
 
Irrelevant whether a A8X exists or not, it's obvious that Apple has to set priorities for the A8 powered iPhones and the iPad Air and mini successors. The higher the demand for smartphones the lower the chances that PRO tablets will ship at the same timeframe.

Else whether A8 or A8X is less relevant; it's more about wafer capacities than anything else probably.
 
af1QuKz.jpg


iPad A8X and removed hardware rotation lock/mute switch. I used the rotation lock multiple times every day. Keep making compromises at the altar of thinness, Apple.
 
The removed switch is just a rumor. We saw a shell with a pilot hole in the right location, so it could very well still be there.

And if it IS removed, it wouldn't be because of thinness, because iPhones are thinner than iPad, and they still have the switch in place. It'd be because of ludicrous Apple cost-cutting, or maybe Jony Ives obsessing over clean design, or something like that. It's not rocket science fitting a switch into a thinner chassis...
 
If the A8X moniker is actually legitimate, that's fodder for a whole new round of speculation! :)
If they do beef up the graphics, I hope they do likewise to the memory interface.

Assuming it is legit:-

I am not totally up to speed regarding Applepay, but is it fundamentally for in-store payments as opposed to online ? If so, then the secure element might get removed in the A8X

Also some of the health related IP might get stripped out, although I assume most of that isn't on the Soc. Does an ipad really need an air-pressure altimeter mostly using for exercise tracking ?

Ultimately, I don't see them putting more CPUs into an A8X, so really the only substantial addition for the SOC can be in the graphics area, i.e. better graphics IP/ higher clock/ better mem bandwidth.
 
The area cost of the secure element is probably a small fraction of a mm2. Probably not much more than a bunch of side band signals and registers between fuses and encryption blocks that can't be read by any CPU.
 
A8X confirmed for iPad Air 2: http://forums.anandtech.com/showthread.php?t=2404005

So this would imply 6-cluster GX6650 GPU with dual core Cyclone-enhanced CPU, and would in turn imply up to ~ 50% higher GPU performance compared to A8 (speaking of which, we will probably not see GX6650 implemented by any other SoC vendor within the next 6 months, which goes to show just how laughably far ahead Apple is compared to any other Series 6XT licensee!).

According to Hans, the mem. bandwidth looks to be 25.6 GB/s, which is the same as Snapdragon 805 and the upcoming Cherry Trail SoC.

Can anyone tell by looking at the picture and/or comparing to A8 what the estimated SoC die size is for A8X?

A8X.jpg
 
Last edited by a moderator:
Just a bit below 32bit K1 area ;)

So this would imply 6-cluster GX6650 GPU with dual core Cyclone-enhanced CPU, and would in turn imply up to ~ 50% higher GPU performance compared to A8 (speaking of which, we will probably not see GX6650 implemented by any other SoC vendor within the next 6 months, which goes to show just how laughably far ahead Apple is compared to any other Series 6XT licensee!).

It could be a theory that with the size Apple has as a partner for IMG they enjoy lead times for example. Even if not who's going to come up with a similar amount of resources to even come close to Apple's integration timings? When the happygomerry rumors started that they're hiring X former GPU team or Y engineers from Lord knows where and therefore are designing their own GPU I had my own theory that Apple needs them primarily in order to minimize as much as possible integration timings. I do not think that it's completely impossible that Apple will come up with any of its own hw designs in the future, but for now all that hired engineering talent have all their hands full with on time integration.

On another note I'd estimate that IMG will most likely announce Series7 before this year runs out. It would be very interesting to see if Apple will skip it next year and for how long :p
 
How can anybody tell the memory bandwidth from that picture??
Well if you read the link...
There are two ELPIDA dram chips next to the A8x and with some imagination we might read the (very blurry) part number right below the ELPIDA logo as "FA164A..." which would make them 16Gb LPDDR3 parts each with a 2x32 bit databus.

That would make a decent 4GB dram in total with a wide 128 bit bus. If I'm reading the "GD-F" on the 3rd line correctly then that should make them DDR3-1600 parts giving the A8X a desktop like bandwidth of 25.6 GB/s.
 
How can anybody tell the memory bandwidth from that picture??

He is making an assumption based on there being two memory chips, each 2x32 wide. Whether his assumtions hold is anybodys guess, but his reasoning is internally consistent. I'm pleased if this holds true, as I felt that this would be necessary to ensure that additional GPU resources are adequately fed compared to its predecessors and real world benefits are fully realized. Bandwidth carries a pricetag, so if we see a return to a 128-bit wide interface for the iPad(s), the case for a migration to TSV-attached memory on 16FF might be strengthened.

Edit: beaten. :)
 
Which should be tempered by the fact that he is also indicating 4GB of ram.
If they really are putting 4 GB of RAM in the iPad Air then I guess multi-app scenarios really are coming as rumoured. Otherwise jumping directly from 1GB to 4GB seems excessive for single app usage or just to address complaints of Safari tab reloading.
 
Well if you read the link...

I still don't see how that would indicate or require the use of a x128 memory bus?, or am I missing something? If it's purely a guess then it would seem to deviate from Apple's recent decisions in terms of power consumption / specs. We'd jump from 1GB to 4GB of RAM, a larger memory bus which you'd expect to increase power consumption. Perhaps an increase in resolution and apparently a reduction the volume of the chassis, so less space for a larger battery, with those conditions Apple would be really doing an amazing job to maintain current iPad battery life. Not long to wait now, to find out the truth :)
 
Back
Top